Abstract: Change trend of mine gas monitoring data is very important to analyze whether the gas is anomaly or not,through sorting out and reviewing the research results at home and abroad,we summarized that the time series of mine gas itself has the tendency,correlation, periodicity and anomalism four characteristics,and also summarized coal and gas outburst,gas after blasting,stopping local ventilation,calibration of sensor probe and sensor probe of gas fault five gas abnormal patterns,and reviewed the grey system,neural network,support vector machine and other methods in prediction of mine gas.Finally, this paper analyzes the mine gas trend of existing problems and shortcomings,and looks forward the K-line theory in the application prospect.